Experimenting and making LTspice models, of 3 BJT regulator and SMPS
I took a 24.2V internal SMPS from a printer, it's a nice enough looking PCB. It's output ripple on the scope is ~7-8mV, with peaks of about 50mV at the switching frequency of 183kHz. And with it passing 0.24A, the overall ripple is still about 8mA, but the peaks are now up around 85mV.
I made a 3 transistor regulator with feedback, that goes from 2.2-22.2V, it only has 1u on input and output.
schematic below
It's no-load ripple is about 50mV. With it at 12.6V @126mA the ripple is like 1.3V. So thats quite the noise amplification.
And when I added a 47u output cap, I get 1.9V (volts not mV) ripple, big solid wavy ripple at 183kHz, not spikes.
In LTspice I made 2 models, 1 with the large scale T-models , and measured the betas of the 3 BJTs and resistors. And the other uses 2n3904 and 2n3055 in spice.
For the input voltage I put a 24V+85mV 183k sine wave. On their output's I get less ripple, only maybe 50uV (micro, not milli like the 80mV on the input). So neither makes the messy thing I made.
Tomorrow I'll hook it up to a 36V transformer, see what it does. As a regulator this thing over compensates under load. The SMPS is fine. I haven't done any calculations yet, this is the 1st time in ages I breadboarded something.